Meta E5
Just cleared the Meta onsite for a Product role. Moving to team match.
Sharing the full breakdown — rounds, questions, what worked, and the resources that actually helped.
YoE : ~ 7 years
⸻
𝗣𝗵𝗼𝗻𝗲 𝗦𝗰𝗿𝗲𝗲𝗻
Before the call, two assessments — a 4-stage coding assessment (cleared 3/4) and a work-style assessment.
The actual screen had two problems:
-
Merge Intervals
→ Given a list of intervals, merge all overlapping ones
→ Sort by start time, then walk through merging where intervals overlap
→ Follow-ups: what if the input is a stream of intervals arriving in real time? How do you handle duplicates? -
Lowest Common Ancestor of a Binary Tree
→ Standard recursive LCA. Missed one edge case where p and q reference the same node — needed a small nudge.
Feedback: strong coding, clear communication, fast implementation. Needed improvement on clarifying questions and edge case coverage upfront.
⸻
𝗢𝗻𝘀𝗶𝘁𝗲 — 𝟰 𝗥𝗼𝘂𝗻𝗱𝘀
Round 1 — Coding
→ Find first 1 in a sorted binary matrix — solved cleanly
→ Trapping Rain Water — solved cleanly
Round 2 — System Design
→ Design the YouTube Live comment system
→ Requirements: millions of concurrent viewers per stream, comments appearing in real time, no strict ordering guarantee, spam filtering, pinned comments by the creator
→ Covered: WebSocket vs SSE for delivery, pub/sub fanout architecture, Kafka for ingestion at scale, Redis sorted sets for real-time comment ordering per stream, CDN offloading for static assets
→ Key tradeoff discussion: exactly-once delivery vs at-most-once at scale — argued for at-most-once with client-side deduplication given the volume
→ Pushed back when the interviewer suggested strong consistency — justified eventual consistency with user experience reasoning
→ Drove the conversation end-to-end
Round 3 — Coding
→ Valid Palindrome II — solved
→ Valid Word Abbreviation — solved
Round 4 — Behavioral
→ Ambiguity handling, conflict resolution, data-driven decisions
→ Questions I remember:
• Tell me about a project you're most proud of
• Tell me about a time something went wrong because of you
• What do you do outside work to improve yourself?
⸻
𝗠𝘆 𝗧𝗮𝗸𝗲𝗮𝘄𝗮𝘆𝘀
-
Clarity > correctness
Meta rewards structured thinking and clear communication over a perfect solution. State your assumptions. Call out edge cases early. They're watching how you think, not just what you produce. -
Ask clarifying questions — always
My phone screen feedback specifically called this out. Asking good questions signals you understand ambiguity, which is a core Meta value. Made it a focus during onsite prep and it showed. -
Hints are normal
Needed a nudge on one problem and still passed. Don't spiral if the interviewer steps in. Other signals matter just as much. -
Pace matters
45-minute rounds move fast. ~5 min intro, ~5 min for your questions. That leaves 35 minutes for two problems. Move with urgency while explaining your thinking.
⸻
𝗥𝗲𝘀𝗼𝘂𝗿𝗰𝗲𝘀 𝘁𝗵𝗮𝘁 𝗵𝗲𝗹𝗽𝗲𝗱
For coding: LeetCode — the Meta tagged problem set is the most focused prep you can do. No shortcuts here.
For system design: Alex Xu's System Design Interview Vol. 1 & 2 are still the best starting point for mental models. Pair with ByteByteGo for visual breakdowns of real systems.
For practicing system design : Prepd (prepd.tech)
⸻
Happy to answer questions on any specific round. Drop them below.
#meta #softwareengineering #systemdesign #leetcode #techinterview #careergrowth


